Seventy-four low birth weight infants underwent serial measurements of cerebral blood flow velocity (CBFV) using Doppler ultrasound together with ultrasound imaging of the brain. Surviving infants were examined and assessed using the Bayley scales of mental and motor development at 18 months post-term. There were no significant differences in mean CBFV between normal and impaired infants. Infants with abnormal neurological signs (n = 6) or developmental delay (n = 3) did not show the usual steady rise in CBFV during the first few days of life. Four of nine had a rise, then a fall in CBFV compared with 4 of 31 with complete data in the normal group. This difference is significant (chi 2 = 4.3, p = 0.03). The magnitude of the percentage increase between day 1 and day 3 was also smaller in the abnormal group (median 0% vs. 39%, Mann-Whitney p = 0.03). A structural abnormality seen on the cranial ultrasound image was a better predictor of adverse outcome than an abnormal CBFV pattern, with a better sensitivity and specificity (66% and 97% for imaging compared to 44% and 87% for CBFV). The addition of serial ultrasound Doppler measurements of CBFV did not improve the prediction of outcome obtained using ultrasound imaging alone.